Inferring the Correlation of Mutation Fitness Effects Between Populations and Gene Functions Wild House Mice

Abstract

Much can be learned about a species’ recent evolutionary past by fitting models to contemporary patterns of genetic variation. One category of model is a demographic history, which provides information about population history. The other main model is a distribution of fitness effects (DFE), which describes the distribution of selection coefficients of new mutations. Using population genomic data from the wild house mouse, Mus musculus domesticus, we used dadi to infer demographic history, and the best model for all pop- ulations pairs was the isolation-migration or the isolation-migration-pre model, both with inbreeding. With the parameters from the demographic inference, we then inferred the joint DFE using all non-synonymous sites for each population pair and found that the best model in all cases was the asymmetric bivariate log- normal distribution. The main parameter of interest in our model is ρ, which quantifies the correlation of fitness effects of new mutations between two populations. For all of our population pairs, we saw a ρ value greater than 0.8 and found that the correlation decreases with genetic divergence, which matches previous results from other species [Huang et al., 2021].To further explore the biological basis of the shape of the joint DFE, we did joint DFE analysis for the Iran and France population pair using only non-synonymous sites annotated to specific gene ontology (GO) and mouse phenotype ontology (MP) terms. In both the phenotype and genotype analysis, we found high correlation of mutation fitness effects between the two populations for all terms. The high correlation we observe at the whole genome level does not seem to change with specific gene or phenotype categories, suggesting that the high correlation of mutation fitness effects is driven by evolutionary forces that are similar in the two populations, such as environment and speciation.